Anthropic cuts cost of its mid-tier model for agent-style work
Reviewed by Helen Jones
Anthropic released Claude Sonnet 5 on June 30. The model reaches performance close to the larger Opus 4.8 on reasoning, tool use, and knowledge tasks while charging $2 per million input tokens and $10 per million output tokens through August 31. The new model is available to all Claude plans and via the API. After the introductory period, pricing rises to $3 per million input tokens and $15 per million output tokens.
Until now, teams avoided building reusable AI workflows because each run felt expensive and the output still needed heavy checking. The lower price removes that cost objection, so the bottleneck moves from model spend to whether anyone has actually defined the context, steps, and review rules that turn one useful chat into a repeatable process. Managers who continue pasting fresh context into every prompt will see their advantage shrink. Those who treat the price cut as permission to codify one recurring task with clear inputs, quality checks, and delegation instructions will capture the real gain.
Analysis
Choose one recurring AI-assisted task and write the workflow down once: the source documents, context, output format, review criteria, and final owner. The cheaper model only helps if the process stops living in one-off chats.
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "Anthropic cuts cost of its mid-tier model for agent-style work", Collab365 Spaces. 1 source referenced.